The accuracy of vehicle speed measured by a speedometer is analysed. The stress on the application of skew normal distribution is laid. The accuracy of measured vehicle speed depends on many error sources: construction of speedometer, measurement method, model inadequacy to real physical process, transferring information signal, external conditions, production process technology etc. The errors of speedometer are analysed in a complex relation to errors of the speed control gauges, whose functionality is based on the Doppler effect. Parameters of the normal distribution and skew normal distribution were applied in the errors analysis. It is shown that the application of maximum permissible errors to control the measuring results of vehicle speed gives paradoxical results when, in the case of skew normal distribution, the standard deviations of higher vehicle speeds are smaller than the standard deviations of lower speeds. In the case of normal distribution a higher speed has a greater standard deviation. For the speed measurements by Doppler speed gauges it is suggested to calculate the vehicle weighted average speed instead of the arithmetic average speed, what will correspond to most real dynamic changes of the vehicle speed parameters.